Verizon’s Community Pulse Survey: The Urgent Case for Accountable AI

Verizon’s Community Pulse Survey: The Urgent Case for Accountable AI

Verizon’s Community Pulse survey tells a more useful story than its enthusiasm for technology suggests. Americans are using AI, small businesses are exploring it across their operations, and public servants see possibilities for emergency response. They are also asking who is responsible when the technology makes a mistake, if employers will help people learn, and what happens to the human relationships that make a community work.

Those concerns are part of the adoption story. Treating them as obstacles to overcome would miss what respondents are saying.

The report, conducted by Morning Consult for Verizon, draws on January–February 2026 surveys of 3,007 U.S. adults, 1,000 small- and midsize-business decision-makers, and 511 public-sector employees. An April survey includes responses from employees, managers, or HR decision-makers. That breadth makes the results useful as a snapshot of attitudes. It does not turn attitudes into evidence that a particular AI deployment improves work, learning, or public safety.

I want to focus on the non-speculative elements of Verizon’s Community Pulse. As I argued in my analysis of the 2026 IBM CEO Study, survey expectations indicate what respondents believe may happen. They do not establish what will happen, much less what will create value. Future-oriented answers can reveal respondents’ current thinking, but without knowing the assumptions behind each choice, they are a shaky basis for planning. For more meaningful data, look at what respondents report from actual experience. Use those findings as a lens for your own experience or as input to marketing, strategy, and product planning. And dig into the numbers before accepting them at face value, as we do below.

Use is widespread. Fluency remains an open question.

Verizon reports that 68% of consumers have used an AI chatbot or assistant. But the figure includes people who use one once a month or less; 39% report using one at least a few times a week (p. 10). The report’s claim of “universal digital fluency” goes further than those answers support. Using a tool says little about whether someone can recognize an error, protect sensitive information, or decide when to exclude it from a decision.

The business figures need similar care. Seventy percent of small businesses report either currently using AI in marketing and social media or planning to do so. Current use is 41% (p. 11). A plan is a statement of intent, not a deployed capability or a customer outcome.

The skills gap is also a management gap.

The April findings are revealing. Only 30% of employees feel very or completely prepared to use AI effectively at work, and 43% say their employer is not providing enough support (pp. 21–22). Verizon says 91% of employers see an AI skills gap, although that total includes the 22% who describe it as small. It also reports that 72% of employees trust employers to help them learn, but half of that group expresses only some trust. The opportunity is real; the depth of confidence is less certain than the headline implies.

Verizon concludes that employers who invest in training “will win” (p. 23). The survey cannot establish that outcome. The more immediate obligation is to give people time, relevant practice, and a way to challenge AI output without being penalized for slowing down a workflow.

In my response to GoTo’s Pulse of Work, I argued that AI adoption has outrun work design. Training people to operate a chatbot while leaving decision rights, review practices, and quality measures untouched will reproduce that gap. Capability develops when people can use AI in the context of their jobs, inspect its work, learn from mistakes, and retain the judgment their organization needs.

Human connection is an operating requirement.

Small business respondents clearly express the tension: 84% agree that AI should supplement employees, while 67% worry that its use would take away the personal connection with customers (p. 12). These answers are compatible. A business can welcome faster service while still recognizing that a relationship suffers when a customer cannot reach someone who understands the situation and has the authority to help.

The same question belongs in education. Sixty-five percent of surveyed parents worry that overreliance on AI could leave their children less prepared for work (p. 17). That is a concern, not evidence that AI has already caused the outcome. It deserves a better response than either a ban or a promise of personalized learning. As I argued in my discussion of learning readiness, the test should be whether AI helps learners develop reflection, judgment, and agency, or lets them bypass the struggle through which those abilities grow. AI is raising deep strategic questions for education that go beyond the scope of this survey response.

Public support gives permission to test, not proof of safety.

Consumers and public-sector employees express support for technologies such as emergency call transcription, first-responder wearables, and AI-based mapping of disaster impact zones (p. 26). Those are reasonable candidates for carefully bounded trials. Yet 80% of public-sector respondents also agree that AI can make mistakes and should not be fully relied upon for emergency or disaster response (p. 12).

A public safety assessment should therefore ask what happens when a map is wrong, a transcription misses a critical detail, a network fails, or an automated recommendation conflicts with what responders see. Who checks the output? Who can override it? Can responders continue without it? In the Serious Insights State of AI 2026 update, I argued for defining what a system may suggest, what it may do, and what it must never do without a human. Emergency services are where those boundaries need to be practiced, not merely promised.

Privacy belongs in the same design discussion. Seventy-seven percent of consumers are concerned about sharing personal information with AI (p. 29). That finding should shape data practices, retention, access, and explanations to the public before any organization asks people to trust a new service.

What Verizon’s next Community Pulse should measure

Verizon has assembled a useful account of public expectations, but the deck leaves important research questions open. While it provides sample sizes and stated margins of error, it does not provide weighting details, response rates, or base sizes for each subgroup shown. Its state highlights also contain apparent copy errors: the Illinois page refers twice to New York respondents (p. 37). State comparisons deserve care when the report itself gives oversample margins of error as wide as 10 percentage points.

A stronger next study would follow the same organizations and communities over time. It would separate planned use from actual deployment, and deployment from results. It would ask employees whether training changed their ability to catch errors; customers whether service improved; and public servants whether a tool made response faster, more accurate, and more resilient when conditions deteriorated. It would publish failures and overrides alongside successes.

The survey’s lasting contribution may be the conditions respondents place on adoption. People are willing to use AI. The survey suggests they are asking institutions to earn that willingness through competence, accountability, privacy, and human support. Those are measurable commitments. Verizon’s next pulse should tell us whether anyone delivered on them.

For more serious insights on AI, click here.

Did you find this analysis: Verizon’s Community Pulse: The Urgent Case for Accountable AI useful? If so, please like, share, or comment. Thank you!

The cover image is AI-generated from the author’s prompt.